Update on the MELA Update on the MELA hypothesis test hypothesis test Nello Bruscino Nello Bruscino Università degli Studi di Napoli “Federico II” & INFN di Napoli Università degli Studi di Napoli “Federico II” & INFN di Napoli th June 2013 11 th June 2013 11 On behalf of the MEGA group: On behalf of the MEGA group: N. Bruscino, F . Cirotto, F . Conventi, C.Dionisi, S. Giagu, G. Gustavino, V . Ippolito, C. Maiani, M. Rescigno, N. Bruscino, F . Cirotto, F . Conventi, C.Dionisi, S. Giagu, G. Gustavino, V . Ippolito, C. Maiani, M. Rescigno, E. Rossi + ANL/UC Chicago E. Rossi + ANL/UC Chicago
Outline Outline Updated Hypothesis tests: Reminder: ● latest results on hypothesis test (MELA) new 2 - h samples: ● recostructed/truth distributions ● WP/RP fraction ● closure test, cos( θ * ) disagreement ● investigate Collins-Soper rest frame partial results Update on the MELA hypothesis test 2
Reminder Reminder Latest results on hypothesis test (MELA) expected/observed CLs change as follows: test CL S (exp) CL S (obs) 0 + vs 0 - 99.8 % 99.4 % 0 + vs 1 + 99.6 % 99.4 % 0 + vs 1 - 99.9 % 95.7 % workspaces published on 0 + vs 2 + 91.4 % 79.3 % / afs/cern.ch/atlas/groups/HSG2/H4l_2 0 + vs 2 + (25% qq) 90.0 % 88.3 % 0 + vs 2 + (50% qq) 90.6 % 93.5 % 013/Moriond/workspaces_spincp/ME 0 + vs 2 + (75% qq) 91.4 % 95.4 % LA 0 + vs 2 + (100% qq) 90.7 % 94.1 % Update on the MELA hypothesis test 3
- New 2 - samples New 2 h samples h added new 2 - - samples: added new 2 h samples: h 2 + + 2 - - 2 - - 2 - - (f qq =100%) 2 2 2 2 h (f qq =100%) h h h m m g 1 =g 5 =1 g 8 =g 9 =1 g 8 =1 g 8 =1 run 4l selection on the 2-h MC f 0 ≠ 0 f 0 =0 f 0 =1 f 0 =0 JHU samples f 1 =0 f 1 =0 f 1 =0 f 1 =1 f 2 ≠ 0 add g i (2 - f 2 =1 f 2 =0 f 2 =0 h ) to build MELA pdf → gg/qq H: pseudo-tensor, g 8 =1 (f 0 =1 or 0, f 1 =0 or 1, f 2 =0) acceptances for both 7 and 8 TeV samples ready Update on the MELA hypothesis test 4
Truth angular distributions Truth angular distributions cos( θ 1 ) cos( θ 2 ) cos( θ * ) JHU paper JHU paper JHU paper 2 -h 2 -h 2 -h MC truth MC truth MC_truth φ 1 φ ✔ MC truth distributions JHU paper JHU paper MC_truth MC_truth are in good agreement 2 -h 2 -h with JHU theoretical paper: 2 -h production OK ! Update on the MELA hypothesis test 5
- New 2 - samples New 2 h samples h 0 + Good-pair and Good-pair and GP WP wrong-pair fractions: wrong-pair fractions: (fraction) (fraction) the kinematics of spin 2 - 8034 817 h 4mu (90.8%) (9.2%) m 1 m 2 involves, downstream of 2m2e the selection, a WP fraction very high compared to 2e2mu “standard” cases. 4240 480 2 -h 4e (89.8%) (10.2%) m 1 m 2 GP WP (fraction) (fraction) Applying the signal selection, the 9498 8166 4mu (53.8%) (46.2%) wrongpair fraction for wrongpair fraction 2m2e 2 - h samples results ≈ 5 2e2mu 50% 0% (“usual” WP fraction). 4109 3787 4e (52.0%) (48.0%) Update on the MELA hypothesis test 6
- New 2 - (f qq =0%) samples:closure test New 2 h (f qq =0%) samples:closure test h First check: closure test closure test for new spin cases for new spin cases First check: φ cos( θ 1 ) Compare pdf(MELA) to MC JHU truth distribution. GOOD agreement in almost all closure tests: θ * cos( θ * ). some discrepancies in cos( ). some discrepancies in Truth selection (2013): Truth selection (2013): cos( θ * ) 50 GeV < m 12 < 106 GeV 50 GeV < m 12 < 106 GeV 12 GeV < m 34 < 115 GeV 12 GeV < m 34 < 115 GeV 115 GeV < m 4l < 130 GeV 115 GeV < m 4l < 130 GeV Update on the MELA hypothesis test 7
- New 2 - (f qq =100%) samples: closure test New 2 h (f qq =100%) samples: closure test h First check: closure test closure test for new spin cases for new spin cases First check: φ 1 cos( θ 2 ) Similarly for 2 - h (f qq =100%) GOOD agreement in almost all closure tests: θ * cos( θ * ), some discrepancies in cos( ), some discrepancies in - smaller than 2 - (f gg =100%). smaller than 2 h (f gg =100%). h p T reweighting p T reweighting only for spin 0 + case! only for spin 0 + case! Truth selection (2013): Truth selection (2013): cos( θ * ) 50 GeV < m 12 < 106 GeV 50 GeV < m 12 < 106 GeV 12 GeV < m 34 < 115 GeV 12 GeV < m 34 < 115 GeV 115 GeV < m 4l < 130 GeV 115 GeV < m 4l < 130 GeV p T reweighting p T reweighting Update on the MELA hypothesis test 8
Truth distributions of p distributions of p T (Higgs) Higgs) Truth T ( Study MC truth distributions in various range of p T : signal events with “high” p of signal events with “high” p T T of the four-lepton system could could the four-lepton system θ * cos( θ * ) ) distribution. warp cos( distribution. warp 6 p T ranger: 6 p T ranger: Select 6 p T ranges (spin 2 - 1. [0.,10] GeV ≈ 10% events h ), with 1. [0.,10] GeV ≈ 10% events 2. [10,20] GeV ≈ 25% events 2. [10,20] GeV ≈ 25% events relative fractions of events. 3. [20,50] GeV ≈ 35% events 3. [20,50] GeV ≈ 35% events 4. [50,100] GeV ≈ 20% events 4. [50,100] GeV ≈ 20% events 5. [100,300] GeV ≈ 10% events 5. [100,300] GeV ≈ 10% events 6. full range 6. full range Update on the MELA hypothesis test 9
θ * cos( θ * ) in the Collins-Soper frame ) in the Collins-Soper frame cos( θ * cos( θ * ) ) is is p p T Higgs dependent : cos( T Higgs dependent a good agreement with pdf(MELA) (i.e. JHU paper distribution) only for p T <20 GeV . To minimize the effect of the transverse momentum, the “alternative” choice is to adopt the Collins–Soper frame (CS frame CS frame): Collins–Soper frame - z-axis (z z CS CS ) = the bisector of the incoming beam momentum and negative of the target momentum in the rest frame of the vector bosons pairs (beam and target not collinear when ≠ , considerably) p T ( Higgs) 0 Update on the MELA hypothesis test 10
θ * cos( θ * ) in the Collins-Soper frame ) in the Collins-Soper frame cos( θ * cos( θ * ) In the Collins-Soper rest frame cos( ) partially loses dependence from partially loses dependence from In the Collins-Soper rest frame p T (Higgs). . p T (Higgs) e e e e m m m m a a a a r r f r f r f d f d r S r S a a C d C d n n a a t t s s 20 GeV 50 GeV good agreement for p T < 20 GeV good agreement for p T < 50 GeV good agreement for p T < good agreement for p T < A good agreement with pdf(MELA) for p T <50 GeV . A good agreement with pdf(MELA) for p T <50 GeV . An excellent agreement with pdf(MELA) for p T <20 GeV . All the other angular variables are not significantly changed (not influenced not influenced by the change of reference frame). by the change of reference frame Update on the MELA hypothesis test 11
θ * cos( θ * ) in Collins-Soper frame ) in Collins-Soper frame cos( All other angular variables are not significantly changed: there's still a good there's still a good agreement. agreement e e m m e e m a cos( θ 1 CS ) cos( θ 2 CS ) m a r r a f a f r S r S f f C C S S C C cos( θ 1 ) e e e e m m m m a a a Φ CS Φ 1 CS a r r r f r f f f S S S S C C C C Update on the MELA hypothesis test 12
C-S rest frame vs. standard rest frame C-S rest frame vs. standard rest frame 2 -h (f qq =0%) The C-S rest frame produces an improvement to understand and improvement reproduce more realistically the new 2 -h samples. 2 -h (f qq =0%) CS rest frame: chi2/ndf = 21/17 Standard frame: chi2/ndf = 30/17 2 -h (f qq =100%) 2 -h (f qq =100%) CS rest frame: chi2/ndf = 7/17 Standard frame: chi2/ndf = 12/17 In each spin case chi2 is improved: there is still a residual discrepancy. However we are further checking... However we are further checking... Update on the MELA hypothesis test 13
Reco distributions: closure test Reco distributions: closure test ) % φ cos( θ 1 ) 0 = f qq φ 1 cos( θ 2 ) cos( θ * ) ( 2 -h ) % 0 φ 0 cos( θ 1 ) 1 = f qq φ 1 cos( θ 2 ) cos( θ * ) ( 2 -h ✔ Reco distributions are in good agreement with pdf(MELA) ⨯Acc(m 1 ,m 2 ,Ω) : closure OK ! Update on the MELA hypothesis test 14
Conclusions ● new 2 -h (qq/gg) samples added, ready for hypothesis testing (produced acceptance fit functions) ● WP fraction in spin case 2 -h is different from the other spin samples: 10% -> 50% 10% -> 50% ● trying to minimize dipendence of cos θ * from p T (Higgs) adopting CS frame ● good agreement (MC vs. pdf) for other angular variables (indipendent from p T ) using CS frame ● improvement for reproducing spin case 2 -h , using Collins-Soper rest frame (improvements for residuals and χ 2 test) Update on the MELA hypothesis test 15
Ongoing work and plans: ● study reco distributions in CS rest frame (spin 2 -h ) ● study truth distributions for other spins: any improvement using CS rest frame? ● implement with latest version of selection ● review and finalization of the workspace (splitting of the theoretical effects, etc) to be more flexible for a possible combination with other channels. Update on the MELA hypothesis test 16
Back up
Recommend
More recommend
